[Valen]

Posting this on behalf of Russell, he's having some difficulty getting the
image links working.

Hi
my first post well here goes
I have a cnc mill that I have converted to servo driven CNC unfortunately
we still need to do manual tool changes. I'm looking at putting an auto
tool changer onto it at some stage, but its on the back burner for the
time being.

The time consuming part of changing tools manually is resetting the
offsets, we thought about using a number of collet holders and just
changing between them, but it felt like the repeatability wouldn't be
particularly good.

So with a tool height gauge we can swap tools in the collet and
automagically measure the offsets, unfortunately they seem pretty highly
priced so I looked at making my own, and saw a number of contact probes
that people had made before (on here and else where) and decided to make
one specifically for tip measurement.

The best part is when we make the automatic tool changer we can keep using
this to take out any variance in the tool holder and also possibly look
for broken tools.
Anyway, here's my first attempt and I'd really like your comments.


It has a steel button moulded to a base using polyurethane(orange colour)
as a spring. Insides and base will be added later.

The insides.


The centre pin just below the wires is the base of the button with a 5mm
thread. The round pcb is epoxied to the inside the top of the body. The
triangle pcb is then pushed down on to the round pcb and tested for all
contacts. The metal triangle backing plate then is bolted to the base of
the button and epoxied to the triangle pcb. Into this is put a base plug
(9mm thick) that slide down onto the lip and protrudes out of the inside
of the body by a 1mm and has a 8mm thread for mounting. In the test one I
bolted it but in the final one this would be epoxyed in and total sealed
Exploded CAD view


did some testing with this set up

By driving the Y axis into the Z post we found the trip range was less
then 0.004 mm over 100 test cycles, it was also getting really cold really
fast while we were doing the test so that might have been affecting it, as
well this is with a 12 volt sense and 4mA current flowing, so there's
probably a fair amount of arcing (well at this scale anyway )
This one need about 1.5 kg to change states, in the final one I was look
at 0.5 kg

In the final version the button will be stainless steel and the body
aluminum along with the the base and a steel 8mm thread stud for
mounting.
The electrics will be gold plated pcb contacts with a better contact
design and circuitry so the voltage across the contact will be around 0.5volts
or less this will reduce any arcing on the contacts, I might have it boost the voltage up to 10 or so when its time to close the contacts again in the switch to help with wetting current/voltage, but being gold there shouldn't be much of an issue.
Also I'll probably fill the thing with argon before its all glued together for that belt and suspenders approach.
As for outputs I'm going to try and be as general as possible, so 5V ttl, differential RS458 say and an open collector output good for say 24v, unless you can think of some others?
Power will be 12V



Any thoughts? comments, suggestions?

[Valen]

forgot to mention we are using glass linear scales for that .004mm measurement so it should hopefully reflect the real world lol

[GaryCorlew]

Looks awsome to me!!

[Valen]

thanks for that, it'll be interesting to see if the home and limit switch versions work as well.

[rusel]

Thank to Valen for posting the start of this thread.
Just did some sums and the average range on the 100 cycle test was 0.0025 mm.
Tell is this any good?

[Fastest1]

Nice work.

[rusel]

Thanks for the encouragement so far.
Have now found a method for the gold contacts, but will let you know if it works

[Valen]

Made a graph up of the first 100 cycles


Initially I figured the drift was due to the temperature changing, now I'm not so sure, I'm wondering if its "wearing in" the contacts or some such, and I'm a little concerned about the large step in the graph as well, though calling .003mm "large" is a new experience for me

We'll have to run the test again at a more constant temperature time of day and see how it goes.

[chris123]

Sounds good
How would an opto couple work as a switch?

[jalessi]

Valen,

What provides tension to the switch?

Jeff...

[chris123]

polyurethane

[Valen]

I looked at optical but they are actually rather poor in terms of accuracy unless you do something like a laser interferomiter. I thaught about it, theoretically you can get really stupidly high levels of accuracy like on the order of 633 nano meters for a full cycle, so you can do even better than that with a bit of mojo. Unfortunatly at that scale things like temperature change let alone vibration will cause a bunch of counts and make picking up the actual trip from the probe much harder. I'm open to suggestions though

[jalessi]

Valen,

What type of polyurethane did you use?

Very nice work.

Thank You

Jeff...

[Valen]

umm a brown one? lol
I'll get russell to give some specifics

[rusel]

Jeff
It is a 55 I can not remember which shore hardness scale though, towards the soft end.

Russell

[rusel]

A bit of a update
We are now looking at putting a 2 sets of LED's that will glow through the polyurethane around the lower edge this will keep them coolant proof. With 4 green when it is powered up and this changes to 4 blue with a 45 degree shift when tool make contact.
Do you think with the colour change and movement of light is a good addition?

Russell

[chris123]

not really if its acts as a limit switch wired to the computer

[Valen]

I was mainly thinking it would be useful during instillation and testing, and it lets you use it without attaching it to the PC.

Having cool glowing stuff on the mill doesn't hurt either ;->

[chris123]

that why in every shop you see undercarriage blue neon light installation kits

[Valen]

I can see it now
"pimp my mill"
we take this 1945 bridgeport and add blue underbody lights, chrome plate the head and supercharge the coolant pump.:rainfro: